NTN TECHNICAL REVIEW No.

75（2007）

[ Technical Paper ]

Development of an Electromechanical Brake

Tatsuya YAMASAKI**
Masaaki EGUCHI**
Yusuke MAKINO**

The NTN electromechanical brake (EMB) system was

designed and launched with the intention of improving overall
performance of future automotive braking systems. NTN has
designed and developed a new linear actuator that can be
applied to this system. This paper reports on the configuration,
design principles, efficiency calculation methods and related
experiments.

is used as a parking brake without any modification, a

1. Introduction
In the automotive industry there has been an
increased emphasis on vehicle safety. Improvement in
brake technology has greatly contributed to stable
running of vehicles. Increased functionality has
resulted in products like ABS, ESC, and brake assist
1)-2). An example of the increased functionality of
automotive brakes is improvement in control
techniques for hydraulic brakes. Furthermore, in an
effort to continue this improvement in functionality and
reduction in environmental impact, automotive
components manufacturers and car manufacturers are
developing electromechanical brake (EMB) systems.
Many of newly developed electromechanical brake
systems employ linear actuators such as ball-screws
and ball-ramps (torque cams). However, when any of
these linear actuators is used to develop the sufficient
thrust required to brake a traveling vehicle, greater
input torque is needed because the load conversion
ratio with the linear actuator alone is insufficient. To
design a more compact, lightweight electromechanical
brake unit, the motor must be more compact.
Therefore an independent reducer needs to be
incorporated. Additionally, the electromechanical
brake is situated in the “unsprung” section of the
vehicle and will be subjected to violent vibration.
Therefore it must be positively fretting-resistant.
Incidentally, when the electromechanical brake unit
separate mechanism will be needed to lock the
revolution motion of the motor because a linear
actuator such as a ball-screw lacks a thrust holding
function. To address this issue and satisfy the
functions required for such an electromechanical
brake unit (high load conversion ratio, fretting-resistant
quality, load-holding function), NTN has invented a
unique linear actuator mechanism that doesn’t require
an accompanying mechanism and developed a
unique electromechanical brake unit that includes an
electric motor.
In this paper, we describe the constitution and
operating principle of this electromechanical brake
unit, the method for calculating the efficiency of the
brake unit, as well as the evaluation tests we have
performed.
Legends
a，b : half width of Hertz contact ellipse on
threaded surface
DS : sun roller outside dia.
DP : planetary roller outside dia.
dBS : ball center diameter on ball-screw
dO : outer ring inside dia.
F : force acting on the piston
k : number of effective thread ridges on each
planetary roller = lP/P
LP : thread lead on planetary roller

**New Product Development Dept. New Product Development R&D Center

**Mechatronics Research Dept. New Product Development R&D Center

-53-
NTN TECHNICAL REVIEW No.75（2007）

LO : thread lead on outer ring

lP : effective length of threaded portion on each
2. Linear actuator
planetary roller 2. 1 Constituent Elements and Operating Principle
n : number of planetary rollers The constitution and operating principle of NTN’s
p : thread pitch linear actuator are discussed below.
F NTN’s linear actuator comprises, as shown in Fig.
PSC : load acting on one screw = ―――――
nk cosβ 1, a sun roller, planetary rollers, an outer ring, a
carrier, a piston and a threaded member which
PN : radial circumferential load from one planetary
F constitutes the screw of the outer ring. Several
roller, acting on sun roller = ――・tanβ planetary rollers are shrink-fitted between the sun
n
roller and the fixed outer ring; thereby the planetary
Pr : circumferential tangential force from one rollers rotate and revolve as the sun roller rotates.
2TS Spiral threading is provided on the outer
planetary roller, acting on sun roller = ＝―――
nDS circumference of each of the planetary rollers as well
Pa : axial tangential force acting on sun roller as on the inner circumference of the outer ring,
TS : sun roller torque wherein these threads have an identical pitch so that
x : axial displacement of piston they can be correctly meshed with each other.
xBS : axial displacement of nut (threaded shaft) on As a result of this layout, while revolving and
ball-screw rotating, the planetary rollers will be axially shifted
WIN : input (workload) relative to the outer ring. The carrier and piston each
WOUT : output (workload) support rotation and revolution of the planetary rollers;
WREV : loss from frictional torque occurring on then, the rotary motion of the sun roller is finally
revolution supporting bearing converted into the linear motion of the piston.
WROT : loss from frictional torque occurring on Suppose that there is no slipping at the contact
rotation supporting bearing points between the outer circumference of sun roller,
WS : loss from frictional torque occurring on sun outer circumferences of planetary rollers and inner
roller supporting bearing circumference of outer ring, the amount of axial
WSC : loss from slipping on screw thread displacement x of the piston relative to the rotary
α : equivalent lead angle motion of the sun roller can be expressed with the
αBS : thread lead angle on ball-screw formula (1). At the same time, the amount of axial
αP : thread lead angle on planetary roller displacement xBS of the nut on the ball-screw can be
αO : thread lead angle on outer ring expressed with the formula (2). The axial
β : thread flank angle on planetary roller displacement x of the piston is dependent on the
δ : amount of offset at contact point on screw difference between the lead angle of the thread on the
thread planetary rollers and that on the outer ring; thereby the
η : efficiency axial displacement x of the piston (our linear actuator)
μ : friction coefficient on circumferential contact can be smaller than the axial displacement xBS of the
surface nut on the ball-screw. In other words, compared with a
μSC : friction coefficient on screw thread value obtained with the ball-screw arrangement, the
θBS : angular displacement of threaded shaft (nut) load conversion ratio, which can be defined as the
on ball-screw thrust relative to the input torque, can be greater with
θS : angular displacement on sun roller our linear actuator.
θREV : angular displacement of revolution on dO
planetary roller＝ ［DS（dO＋DS）］ ・θS x＝――・（tanαO−tanαP）
・θREV …………(1)
2
θROT : angular displacement of rotation on
dBS
planetary roller＝ ［−DS（dO−DS）］ ・θS xBS＝――・tanαBS・θBS………………………(2)
2
θ'ROT : angular displacement of rotation on planetary
roller (relative to carrier)＝θROT−θREV
Also, we have attempted to improve fretting
resistance of our electromechanical brake by applying
a negative clearance preload in the radial direction
among the sun roller, planetary rollers and outer ring.
To achieve, at a lower cost, both adjustment of radial
clearance and formation of thread, the thread is

-54-

Threaded member
Planetary roller
Forward Rotation
displacement
Sun roller
Piston
Carrier
Outer ring
Fig. 1 Schematic of linear actuator
formed on the outer ring by adding a coil-shaped
thread member in the spiral grooves formed on the
bore surface of the outer ring.
2. 2 Efficiency
In order to be able to determine the efficiency of our
linear actuator, we provided the following assumptions
[1] through [4]:
[1] The thrust occurring on the piston is uniformly
carried by all the planetary rollers and threads.
[2] Because of the smaller lead angle (not greater than
5 degrees), the circumferential component of the
load occurring on the thread can be ignored.
[3] No slipping occurs at the contact points between
the outer circumference of the sun roller and the
outer circumference of each planetary roller and
between the inner circumference of the outer ring
and the outer circumference of each planetary roller.
[4] The areas where loss appears to occur are the
revolution supporting bearing, rotation-supporting
bearing, and sun roller supporting bearing and
contact surfaces on threads.
In addition to these areas, the axial sliding at the
contact areas between the planetary rollers and sun
roller and between the planetary rollers and outer ring
is regarded as a loss. However, this sliding
accompanies rolling contact, and is considered to be
fairly small.
Rotation supporting bearing
Revolution
supporting
bearing
Thrust：F Sun roller Torque：TS
Axial supporting bearing Angular
displacement：x displacement：θS
Fig. 2 Relation of input and output
Development of an Electromechanical Brake
By referring to Fig. 2, suppose that the torque being
input from the sun roller causes the piston to move
axially. Then, the efficiency of the linear actuatorηand
the relation between the work input and work output
can be expressed with the formulas (3) and (4),
respectively.
WOUT F・x
η＝―――＝――― …………………………(3)
WIN TS・θS
WOUT＝WIN−WREV−WROT−WS−WSC ………(4)
The torque loss from each bearing was calculated
utilizing the Palmgren 3). experimental formula for
friction torque.
Now, we would like to discuss the load acting on the
sun roller supporting bearing. The normal force and
tangential force from a planetary roller, shown in Fig.
3, act on the sun roller. From the relation with a
frictional circle, we determined the axial tangential
force Pa with the formula (5) below, and took the
resultant force value as the axial load acting on the
sun roller supporting bearing.
Pa＝n √ 2
（μPN）−Pr2 ………………………(5)
PN
Pr＝2TS/nDS
μPN
Pa/n
Fig. 3 Load on sun roller
Next, we would like to explain the losses arising
from sliding on the screw thread. On the screw
threads between the planetary rollers and outer ring,
the Hertz contact ellipse shifts, as illustrated in Fig. 4,
while the planetary rollers rotate. Suppose that each
Housing planetary roller moves along the inner circumference
of the outer ring. Then, the contact ellipse on the
screw thread shifts with the sliding motion that
consists of the tangential displacement U of the center
of ellipse (expressed with the formula (6)) and the
rotational displacement θ on the center of ellipse
(expressed with the formula (7)).
dO DP
（ 2 ）
U＝ ――−δ ・θREV＋ ――−δ ・
2 （
（θROT−θREV） ）
＝δθROT ……………………………………(6)
（θROT−θREV）
θ＝ ・COSβ＝θ'ROT・COSβ ……(7)
-55-
 NTN TECHNICAL REVIEW No.75（2007）

Then, take the sliding-induced losses within the Under the calculation conditions given in Table 1,
contact ellipse as the sum of losses within a miniature the thrust F and efficiencyηfor the units No. 1 through
area such as a one given in Fig. 5. Then, from the 4 were calculated, and the result is plotted in Fig. 6;
amount of sliding V (x, y), determined with the formula and the loss ratios relative to the inputs to the unit No.
(8) and the normal load f (x, y), determined with the 1 are illustrated in Fig. 7. For these calculations, it
formula (9), the sliding-induced loss WSC on the screw was assumed that μ=μSC = 0.1. From Fig. 6, it
thread can be expressed with the formula (10). should be understood that the thread can be variously
specified and a variety of linear actuators each
V(x, y)＝ (U＋rθsinφ)2＋r 2θ2 cos2φ ……… (8) featuring unique efficiency can be designed. With Fig.
7, it is possible that the proportion of the loss resulting
3PSC x2 y2 from sliding on the screw thread is large and
f (x, y)＝ 1− − ・dx・dy …… (9)
2πab a2 b2 increases as the torque increases.

WSC＝nk∫μSC・f(x, y)・V(x, y)…………… (10)

Table 1 Calculation condition

Technical data No.1 No.2 No.3 No.4
Outer ring
DS, mm 12 ← ← ←
DP, mm 30 ← ← ←

Contact ellipse dO, mm 72 ← ← ←

δ n 3 ← ← ←
LP, mm 2 2 0 3
LO, mm 4 2 2 6
p, mm 2 ← ← 3
IP, mm 16 ← ← ←
δ, mm 0.3 ← ← ←
Screw thread

Planetary roller 40 80

Fig. 4 Contact ellipse in thread of the screw

30 60

Efficiency %
Thrust kN

x dx
20 40

U
dy 10 Thrust Efficiency
No. 1 20
r No. 2
y No. 3
φ No. 4
0 0
0 1 2 3
rθ V(x.y)
Torque Nm

Fig. 6 Calculation results of thrust and efficiency

Contact ellipse
Fig. 5 Contact ellipse
80
Revolution supporting bearing
Rotation supporting bearing
Sun roller supporting bearing
60 Screw thread
Loss ratio %

40

20

0
0 1 2 3
Torque Nm

Fig. 7 Ratio of losses

-56-
 Development of an Electromechanical Brake

3. 2 Efficiency and load holding function

3. Evaluation for characteristics of The test results of thrust and efficiency for the units
linear actuators A and B are given in Fig. 9. These results coincide
3. 1 Units being tested and test method with the result of simulated calculation. This means
We have evaluated the efficiency and thrust holding our calculation practice is reliable.
function of our linear actuator. The specifications of The relation between the torque and thrust on the
the linear actuator units tested are summarized in linear actuator in one cycle (a duration where the
Table 2, and the configuration of the test rig is thrust is increased from a zero level to a particular
illustrated in Fig. 8. With the test rig, torque was level, and then is decreased to a zero level) is shown
applied to the linear actuator by, for example, applying in Fig. 10. When the thrust is decreased, a negative
a weight to the lever, thereby the efficiency of our torque is necessary on the unit A while a positive
linear actuator was determined based on the torque torque occurs on the unit B. This is because the sun
input to the sun roller and the thrust from the piston roller rotates by the thrust and, consequently, its thrust
detected by the load cell. holding function is lost.
To address this issue, we developed the relationship
between the equivalent lead angle of the linear
actuator defined with the formula (11), efficiency of the
linear actuator when it develops 30 kN and availability
Table 2 Specifications of linear actuator
of its thrust holding function, and this relation is plotted
Unit A B C D E F
in Fig. 11. With a greater equivalent lead angle, the
DS, mm 12 8 15 ← ← ←
efficiency is better; when the equivalent lead angle is
DP, mm 30 26 28.5 ← ← ← 0.331 deg or greater, the thrust holding function is not
dO, mm 72 60 72 ← ← ← available. To sum up, it must be understood that the
n 3 ← 4 ← ← ← linear actuator needs to be designed to have an
LP, mm 2 0 2 0 2 3 equivalent lead angle of approximately 0.3 deg so that
LO, mm 4 2 4 2 2 6 it can have the thrust holding function.
p, mm 2 ← ← ← ← 3 2x
α, deg 0.173 0.536 0.220 0.419 0.640 0.331 （
α＝tan-1 ――――
DS・θS ） ………………………(11)
Load cell Test unit Torque meter Lever

Fig. 8 Test rig for measuring efficiency

40 80 40 80

30 60 30 60
Efficiency %

Efficiency %
Thrust kN

Thrust kN

20 40 20 40

10 20 10 20
Thrust Efficiency
Thrust Efficiency
Test Test
calculation calculation
0 0 0 0
0 1 2 3 0 1 2 3
Torque Nm Torque Nm
(a)Unit A (b)Unit B
Fig. 9 Test results of thrust and efficiency

-57-
NTN TECHNICAL REVIEW No.75（2007）

40 40

30 30
Thrust kN

Thrust kN
20 20

10
10

-1 0 1 2 3 4 -1 0 1 2 3 4
Torque Nm Torque Nm
(a)Unit A (b)Unit B

Fig. 10 Relation between torque and thrust

80 4. Evaluation for electromechanical

brake units
60 4. 1 Electromechanical brake units tested
We have manufactured a prototype
Efficiency %

electromechanical brake unit that can be incorporated

40 into an actual car. It has been evaluated on a test rig
and tested as mounted on an actual car. The
20 specifications of the electromechanical brake units
and motor used are summarized in Tables 3 and 4,
Load holding function is valid.
Load holding function is not valid.
while the configuration of the electromechanical brake
0 (EMB) unit is illustrated in Fig. 12 and that of the
0 0.2 0.4 0.6 0.8
Equivalent lead angle deg.
motor is given in Fig. 13. So that the axial length of
EMB unit can be shorter, the motor is arranged
Fig. 11 Relationship between equivalent lead angle, parallel with the linear actuator and the motor itself is
efficiency and load holding an axial clearance-type motor that has two stators.
From this motor, the torque is transmitted via a gearing
to the sun roller. For convenience of incorporation into
an actual car, the linear actuator used is the type B unit
described in Sec. 3, which features a compact size.

Pad Linear actuator X

Y

Motor

Y X

SECTION X-X SECTION Y-Y

Brake rotor
Fig. 12 Prototype EMB unit

-58-
 Development of an Electromechanical Brake

Table 3 Specifications of EMB units Rotor Stator

Characteristic Specification
Max. thrust 30kN
Gear
Size (excluding caliper) 150mm×82mm×85mm
Mass (excluding caliper) 3.8kg

Table 4 Specifications of motor

Characteristic Specification
Max. output 140W
Y
Max. torque 2Nm
Max. speed 1500min-1

Fig. 13 Axial gap motor

4. 2 Evaluation about response
On the test rig illustrated in Fig. 14, we evaluated
the response time relative to thrust on our EMB unit
(increase and decrease characteristics). The thrust
was measured with a load cell that was mounted in
place of a brake rotor and pad. The test conditions are
summarized in Table 5, and the test results are
plotted in Figs. 15 and 16.
In our test project, a particular voltage was fed into
the motor 0.1 sec. after the beginning of each
measuring run, thereby the thrust increase/decrease
characteristics were evaluated. Under any
combination of test conditions, the time-dependent
variation in the thrust is nearly constant in the region
approximately 10 kN or greater. In the thrust range 10
kN or lower, this variation does not
appear to be linear. It is possible, by applying a
greater voltage to the motor, to promote the variation
This is because with a greater voltage applied, the
maximum motor speed is greater.
Next, we want to discuss the effect of an initial
clearance between the piston and load cell. Let us
think of the cases where the initial clearance is set to
0.2 mm. When the voltage applied is 12 V, the time
needed for a thrust to occur takes approximately 0.1
second while this time span is as short as
approximately 0.05 second when the voltage is 20 V.
Thus, the size of initial clearance significantly affects
the response time.
Table 5 Test condition
Thrust
Parameters
Increase characteristic Decrease characteristic

in thrust, in either a thrust increase or decrease mode, Voltage applied 12V，20V

to shorten the time needed to reach a targeted load. Initial clearance 0mm，0.2mm ―

Piston
Load cell Linear actuator
Motor

Fig. 14 Test rig for measuring the response time

-59-
 NTN TECHNICAL REVIEW No.75（2007）
Thrust kN, Current A, Voltage applied V

Thrust kN, Current A, Voltage applied V

60 3000 60 3000
Thrust Thrust
Speed Speed
50 Current 50 Current
Voltage applied Voltage applied

Speed min-1

Speed min-1
40 2000 40 2000

30 30

20 1000 20 1000

10 10

0 0 0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Time sec Time sec
(a) Voltage applied 12V, clearance 0mm (b) Voltage applied 20V, clearance 0mm
Thrust kN, Current A, Voltage applied V

Thrust kN, Current A, Voltage applied V

60 3000 60 3000
Thrust Thrust
Speed Speed
50 Current 50 Current
Voltage applied Voltage applied
Speed min-1

Speed min-1
40 2000 40 2000

30 30

20 1000 20 1000

10 10

0 0 0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Time sec Time sec
(c) Voltage applied 12V, clearance 0.2mm (d) Voltage applied 20V, clearance 0.2mm

Fig. 15 Response time (increase characteristic)

Thrust kN, Current A, Voltage applied V

Thrust kN, Current A, Voltage applied V

60 3000 60 3000
Thrust Thrust
Speed Speed
50 Current 50 Current
Voltage applied Voltage applied
Speed min-1

Speed min-1
40 2000 40 2000

30 30

20 1000 20 1000

10 10

0 0 0 0
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4 0.5 0.6
Time sec Time sec
(a) Voltage applied 12V (b) Voltage applied 20V

Fig. 16 Response time (decrease characteristic)

4.3 Test on actual car configuration shown in Fig. 18, a thrust value detected
As illustrated in Fig. 17, we incorporated our by the strain gage relative to a command value from
prototype EMB unit into each of rear wheels on the the controller is fed back to the controller in order to
test car. Our EMB unit could be readily installed to the control the motors. We attempted to brake the car
test car only with adjustment to the mounting position, traveling at a speed of approximately 30 km/h, by only
and without modification to the area including the actuating the EMB units on the rear wheels, and could
knuckle. Also, we have already verified that our EMB reliably stop the car. For this test, we used a 12-V
unit can be incorporated into the front wheels of the battery for automotive applications.
car without any significant modification. On the system

-60-
 Development of an Electromechanical Brake

(EMB) unit that includes our linear actuator and an

electric motor, and obtained the following results:
(1) We have developed an efficiency calculation
technique for our linear actuator, and verified the
validity of this technique through a series of tests.
(2) We have measured the thrust and efficiency of our
linear actuator, thereby we have clarified the
thread specifications for satisfying the thrust
holding function.
(3) We have learned the characteristics of the EMB
unit (a unit including an electric motor) through a
Fig. 17 Installation of the unit
response evaluation test and a braking test on an
actual car.

5. Conclusion
NTN has invented a unique linear actuator for EMB
units that will be marketed in the near future, wherein
our linear actuator, though not having any additional
mechanism, not only boasts a high level of load
conversion ratio and high degree of fretting resistance
but also features a unique load holding function.
Then, we have evaluated an electromechanical brake
References
1) Tomohiko ADACHI: The Electronic Stability Control
System, Journal of Society of Automotive Engineers of
Japan, Vol.60, No.12 pp.28-33 (2006)
2) Bo Cheng, Tetsuo Taniguchi, Tadashi Hatano, Toshiya
Hirose: Effect of Brake Assistance System in
Emergency Situation, Society of Automotive Engineers
of Japan, Lecture Session Preprints, No.20065894
3) A. Palmgren：Ball and Roller Bearing Engineering,
3rd ed., Burbank, Philadelphia, (1959) 34.

Hydraulic brake Motor driving battery Strain gage

Command Electromechanical brake unit

Motor
Inverter
phase
Inverter driving
battery
AC（U, V, W）
Controller Inverter

Command Electromechanical brake unit

Hydraulic brake Motor driving battery Strain gage

Fig. 18 Prototype EMB system diagram

Photos of authors

Tatsuya YAMASAKI Masaaki EGUCHI Yusuke MAKINO

New Product Development Dept. New Product Development Dept. Mechatronics Research Dept.
New Product Development R&D New Product Development R&D New Product Development R&D
Center Center Center

-61-